Every sales page about Dholera has a risk section, and every one I have read follows the same recipe. Three soft risks are listed, and each resolves, within a sentence or two, into a reason to buy sooner. Execution takes time in India, which is why early entry is rewarded. Infrastructure is still coming up, which is why prices are still low. Approvals are slow, which is why patient capital wins. That is not a risk list. That is an objection handled in advance, wearing the costume of candour.
This page is the version I would want if the money were mine. It is the ledger, unhedged, with dates attached, and nothing written to be reassuring. Eight risks, plus one cautionary precedent that belongs on every Dholera investor's desk, and after each entry the honest answer to what a buyer can actually do about it, including the entries where the answer is nothing.
Two things before the first line, because they set the tone. I think Dholera is the most serious greenfield industrial project in India right now, and I have argued that at length elsewhere on this site. I also think the case for it is strong enough that it does not need help from a padded risk section. If an argument only survives when the risks are softened, it was never an argument.
Risk one: the whole thesis leans on one factory
Strip away the renders and Dholera's investment case rests on a single asset: the Tata Electronics and PSMC semiconductor fabrication plant, approved by the Union Cabinet on 29 February 2024 under the India Semiconductor Mission, carrying Rs 91,000 crore of investment, with the fiscal support agreement signed on 5 March 2025 citing Rs 91,526 crore. At maturity it is designed for up to 50,000 wafers a month on 300 mm wafers, across nodes of 110, 90, 55, 40 and 28 nanometres, making power management chips, display drivers, microcontrollers and logic. Civil construction was reported past the halfway mark by mid-2026 with cleanroom fit-out underway. A 66 hectare Tata Semiconductor special economic zone was notified around April 2026, with roughly 21,000 jobs projected in reporting. First silicon is targeted around December 2026 and commercial production is reported for mid-2028.
Now the sentence the brochures leave out: no chip has been produced there yet. Everything above is approval, agreement, construction and target. Those are real and they are not nothing, but they are not output, and until a wafer comes off that line the single most important economic fact about Dholera remains unproven.
Then look at the concentration. The fab's own employment claim is more than 20,000 jobs, direct and indirect. The promotional city speaks of roughly 800,000 jobs at maturity, which official material breaks into about 312,900 direct and about 483,630 indirect. The anchor everyone is buying into therefore accounts for something on the order of a fortieth of the promised employment. The rest depends on the electronics, pharmaceutical, heavy engineering, automotive, general manufacturing, agro and IT tenants named in the sanctioned plan turning up in strength over two decades. They might. Today, one of them is pouring concrete and the others are prospects. I have made the fuller version of this argument in why the fab is the ballgame, and the phrase is not affection, it is exposure.
What you can do about it: very little inside Dholera, which is the point. You cannot diversify within a single anchor region the way you can across a portfolio, so this is a position sizing question rather than a diligence question. Decide the number you can leave in the ground for a decade with no yield, and let the concentration set that number rather than your enthusiasm. The one structural comfort is that the trunk infrastructure, the transferred land and the transport links are not fab specific. A fab failure would gut the story and the sentiment. It would not unbuild the roads.
Risk two: this place has a documented relationship with its own deadlines
Dholera does not miss dates occasionally. It misses them as a pattern, and the pattern is old enough to be predictive.
The airport is the cleanest example. Targets for Dholera International Airport have slipped repeatedly since around 2010, and December 2025 came and went without operations. As of July 2026 it stood at roughly 80 per cent complete, with the runway, taxiways and air traffic control reported finished and the terminal at about 75 per cent, and operations targeted for September or October 2026. A trial and calibration landing by an Airports Authority of India aircraft took place on 4 June 2026, aircraft VT-CNS. That landing proves the runway works. It does not make the airport open, and I will keep saying so until a scheduled passenger flight departs. The way to read these announcements without being played is set out in how to read the airport dates.
The activation area carried a programme of roughly 120,000 residents and 80,000 jobs by 2020. That target lapsed unmet, and no honest account of Dholera can leave it out. The remaining 700 MW of the solar park's sanctioned 1,000 MW Phase I is targeted for March 2027 and has been delayed by tariff and regulatory disputes at the state commission. The semi-high-speed rail line approved on 13 May 2026 carries a completion target running up to 2030-31, which is five years of weather ahead of it. Even the master plan's own end year is unsettled: some official documents show 2040, others 2042. I flag that discrepancy rather than resolve it, because I cannot source a resolution, and a project whose finish line moves by two years on paper is telling you something about its calendar.
The counterweight is real and I will not bury it. The Ahmedabad to Dholera expressway, about 109 km of access-controlled greenfield road, was reported inaugurated on 31 March 2026 and is operational. The NICDC Delivery Monitoring Unit report to DPIIT dated 30 June 2026 records the activation area's trunk infrastructure works as complete. Things do land here. They land late, and the gap between the announced date and the actual date has been the single most reliable feature of this project for fifteen years.
What you can do about it: underwrite on delivered items only, and treat every target as a scenario rather than a schedule. If your purchase makes sense only when the airport opens this year, or only when the fab ships in December, you have not bought land, you have bought a deadline, and this place eats deadlines.
Risk three: the ground itself is a constraint
Dholera sits on the Bhal, the flat low-lying coastal belt along the Gulf of Khambhat, in a semi-arid climate. Flooding and drainage are not activist talking points here, they are documented constraints inside the planning literature, which is why so much of the engineering effort has gone into water at all. Roughly one third of the developable area falls within the Coastal Regulation Zone, a figure I hold as reported government material rather than durable, and CRZ land is not a technicality. It is a legal regime that governs what may be built and where.
The engineered answer, as recorded in the sanctioned plan and the DICDL account of the built utilities, is genuinely substantial: a 50 MLD water treatment plant, about 100 MLD of potable water available, a 10 ML reservoir, 82 km of water pipeline with smart metering and non-revenue water reported under 5 per cent, a 10 MLD sewage treatment plant, a 20 MLD common effluent treatment plant, 81 km of recycled water pipeline and stormwater handled through a 6.5 km canal. That is a serious water system by Indian urban standards, and dismissing it would be as lazy as swallowing it whole.
The honest framing is a three-way split. Some of this is solved, in the sense that it exists and functions at the activation area's current scale, which is a scale with almost no residents drawing on it. Some of it is designed, meaning it appears in the sanctioned plan for a city of about a million people that does not exist. And some of it is unproven, because no coastal Indian city of this ambition has yet demonstrated that this design survives a bad monsoon, a full industrial load, and two decades of maintenance politics simultaneously. Water in a semi-arid coastal region supporting heavy industry is the kind of problem that looks solved on paper for years and then does not. I have taken the whole question apart in the water and climate essay.
What you can do about it: find out which zone your specific parcel sits in before you find it out afterwards. The plan's zone types include industrial, residential, City Centre, High Access Corridor, Knowledge and IT, Logistics, Strategic Infrastructure, Public Facilities, sports and recreation, solar park, green belt, CRZ, agriculture and village buffer. A parcel in the CRZ or a green belt is a different asset from a parcel in a residential zone, whatever the brochure calls it, and the per zone percentages that sellers sometimes quote are not something I have seen reliably sourced.
Risk four: the land history is litigated, and the paperwork risk is entirely yours
Dholera's land story did not begin smoothly. The Gujarat High Court stayed parts of the SIR acquisition in 2015 after farmer petitions. In 2017, Business Standard reported that only about 290 of the 900 plus sq km had then been secured, in the same period when the wider Delhi Mumbai Industrial Corridor programme was being described in print as a tale of abandonments and delays. That is the base from which today's progress was built, and the DMU report now records 48.31 sq km transferred to DICDL, which is both real progress and a reminder of how much of the envelope is still not in the builder's hands.
The retail buyer inherits a smaller, sharper version of the same problem. The documented red flags in this market are consistent: agricultural land marketed as being inside the SIR, plots sold at distances from the activation area that the map does not support, schemes that are not registered, and title that nobody has traced. Gujarat exempts some plot-only schemes from RERA registration, and where a seller claims that exemption the burden does not disappear, it shifts entirely onto your title diligence. Even the units can be slippery: a bigha in Gujarat is commonly treated as about 2,500 square yards but is not standardised, while an acre is fixed at 4,840 square yards or 43,560 square feet, so a quoted rate per bigha can mean different quantities of earth depending on who is quoting it.
What you can do about it: this is the one major risk on the ledger that sits almost fully within your control, which is why I get impatient when buyers obsess over macro forecasts and skip it. Verify the scheme's GUJRERA registration number and status on the state portal, and insist on clear, traced title inside the SIR before any money moves. That means the mother deed and the chain of ownership, a 30-year encumbrance certificate, the 7/12 extract, tax receipts, biometric registration at the sub-registrar and a Khata in your name. A seller who resists that sequence has told you something more useful than any brochure. The five minute version of the first check is in the GUJRERA check. One further rule for overseas buyers: non-resident Indians and OCI holders may buy residential and commercial property but not agricultural land, which makes the agricultural land mislabelling risk considerably more dangerous for them than for a domestic buyer.
Risk five: the people have not arrived, and people are the hard part
The 2011 census counted 2,779 people in Dholera village, 1,420 men and 1,359 women, across 576 households, with literacy at 80.29 per cent. That is the demographic base. The promotional city speaks of roughly a million residents at maturity. Between those two numbers lies the entire risk of this project, and no amount of completed pipework closes it.
I keep returning to this because Indian urban history is unambiguous on the point: infrastructure is the easy half. This country knows how to lay roads, string power and build treatment plants. What no plan has ever been able to schedule is the arrival of households, and a road compels nobody to live beside it. Dholera has already failed this test once, at the 2020 activation target, and the absence of residents is not a cosmetic issue. It means there is no local retail economy, no rental market to speak of, no service ecosystem, no schools and hospitals at city scale, and therefore no organic demand pulling land values along independently of announcements. Every valuation argument being made about Dholera today is an argument about the future, because the present does not contain enough people to make one.
What you can do about it: nothing. This is not a risk you diligence away, it is the bet itself, and anyone selling you Dholera without saying so plainly is selling you a version of the place that does not exist yet. The only rational response is horizon. If you cannot wait through the years it takes for households to follow employment, and employment to follow a factory that has not yet shipped a chip, then the asset is wrong for you regardless of how the paperwork looks.
Risk six: you cannot see the price, and neither can the person quoting you one
There are no reliable public per-unit prices for land in Dholera. I want that sentence to land properly, because it is the least discussed and most expensive risk on this page. Figures do circulate, usually per square yard, usually delivered with great confidence, and I am not going to repeat one here or anywhere else on this site. They are broker-tier claims, and a claim is not a price.
The reasons are structural rather than conspiratorial. Transaction volumes are thin, so there is no dense record of comparable sales to average. There is no public listing system that verifies what actually changed hands. The government valuation grid and the negotiated market number are different animals in every Indian district, and Dholera is no exception. The units themselves are ambiguous, since a bigha is not standardised. And most decisively, the same district contains parcels that are functionally different assets: land inside the activation area with trunk services at the boundary, land inside a sanctioned town planning scheme but far from services, and land that is agricultural in everything but the sales pitch. A single rupees per square yard number covering that spread is not information, it is decoration.
The consequence is that you cannot mark your holding to market, you cannot audit an appreciation claim someone shows you, and you cannot know with confidence whether you overpaid until you try to sell. Add the friction that is knowable: stamp duty at an effective 4.9 per cent, being 3.5 per cent plus a 1.4 per cent surcharge, plus 1 per cent registration, and brokerage in both directions, against an asset that produces no income while you hold it. I will not print an expected return, because no honest series exists from which to compute one, and any percentage offered to you was manufactured rather than measured. The related seller vocabulary deserves the same treatment: assured returns and guaranteed appreciation are marketing language, not financial facts, and in a market with no verifiable prices they are unfalsifiable by design.
What you can do about it: treat every number you are shown as a claim to be tested rather than data to be absorbed. Registered sale deeds are public records at the sub-registrar, so a lawyer can triangulate what comparable parcels actually registered at, which is closer to evidence than anything on a brochure. Then ask the seller what their number is derived from. The quality of that answer is itself a data point.
Risk seven: the pipeline is not the same thing as installed capital
Dholera's investment headlines mix at least three tiers of certainty, and sellers rarely separate them, so I will.
Tier one is signed and building. The fab has Cabinet approval, a signed fiscal support agreement and civil work reported past halfway. The expressway is reported open. About 300 MW of the solar park is commissioned by Tata Power, using 873,012 modules across 1,320 acres, verified through Central Electricity Authority reporting of September 2025 and the DMU report. The activation area's trunk works are recorded complete. This tier is the real balance sheet.
Tier two is agreements and memoranda, which are intentions with letterhead. The L and T Vyoma artificial intelligence data centre agreement is confirmed through a government news release dated 20 February 2026, but the Rs 25,000 crore and 250 MW figures attached to it are memorandum stage, with operations spoken of around 2028. Fujifilm India signed a memorandum with the Gujarat State Electronics Mission on 30 June 2026 to explore a semiconductor materials base, and explore is the operative verb. Embraer and Adani have a memorandum exploring a final assembly line around 2028 with no site finalised. Tsingshan's steel and battery investment of roughly Rs 21,000 crore sits at reported tier. Two further Cabinet cleared semiconductor units for Gujarat, one at Dholera and one at Surat, together carry more than Rs 3,900 crore and more than 2,200 jobs in May 2026 reporting, which I hold as amber.
Tier three is state policy ambition, which is not Dholera capital at all. The Viksit Gujarat data centre policy for 2026 to 2029 carries a state-wide ambition of Rs 6 lakh crore and 7.5 GW with Dholera named as a primary cluster. That is a target for a state, not committed money for a district, and quoting it as Dholera's investment is one of the most common sleights I encounter. The same applies to India's semiconductor programme overall, roughly a dozen sanctioned projects worth about Rs 1.64 lakh crore nationally, which belongs to the country and not to this taluka. Even the cumulative pipeline figure I am willing to cite, more than Rs 1.5 lakh crore of confirmed private investment across the Dholera and NICDC pipeline, is a pipeline number rather than a Dholera number, and I hold it amber for that reason.
What you can do about it: before any figure moves you, ask which tier it belongs to. Signed and building, memorandum stage, or state ambition. That single question deflates most of the excitement in this market, and what survives it is worth taking seriously.
Risk eight: the exit is the risk nobody rehearses
Buyers spend months on entry and almost no time on exit, which is strange, because the exit is where the entire return is realised in an asset that pays nothing while you hold it. There is no reliable public resale data for Dholera land, so nobody can honestly tell you how long a sale takes or at what discount to the asking number it clears. The mechanics are not mysterious though. Your buyer, some years from now, will run exactly the checks you should be running today, and every gap you tolerated at purchase becomes a discount you absorb at sale.
That gives illiquidity a shape you can work with. A parcel inside the boundary, in a registered scheme, with traced title, unambiguous measurements and a clean registration record is the version of this asset that can be transacted at all. A parcel with a fuzzy boundary, a plot-only exemption claim, an unresolved chain of ownership and a bigha measurement nobody can reconcile is not cheap, it is difficult, and difficulty is priced by the person on the other side of the table. I have set out the mechanics in the resale market reality.
What you can do about it: buy the version that is easy to sell, and accept that this discipline costs you something at entry. It is the closest thing to liquidity insurance available in this market.
The precedent that belongs on every Dholera desk
In 2022, a Vedanta and Foxconn joint venture announced semiconductor manufacturing in Gujarat with a headline value of about 19.5 billion US dollars. It was covered as transformative. Land was discussed, timelines were floated, and the announcement did what announcements do to sentiment. On 10 July 2023, Foxconn withdrew, and the venture dissolved without a plant being built.
I raise it not to sneer, because failed ventures are ordinary in heavy industry, but because it is the cleanest available lesson in tier discipline. An announced anchor is not an anchor. A headline capital figure attached to a memorandum can go to zero without anything visible happening, and anyone who bought land on the strength of that headline was holding a position underwritten by a press release.
The Tata project is categorically further along, and I say that as plainly as I say the rest: Cabinet approval on 29 February 2024, a fiscal support agreement signed on 5 March 2025, construction reported past the halfway mark, a special economic zone notified around April 2026, and a partnership with ASML confirmed through the Tata Electronics newsroom. That is not a memorandum. It is a plant being built. It is also not yet a chip, and the discipline that made the Vedanta and Foxconn collapse survivable for careful investors is the same discipline that applies here.
The ledger in one table
Everything above, compressed. Where the mitigation column says none, I mean it literally, and I would rather write the word than invent a comfort.
| Risk | What the record shows | What you can actually do |
|---|---|---|
| Anchor concentration | Rs 91,000 crore fab approved 29 February 2024, first silicon targeted around December 2026, no chip produced yet; fab jobs above 20,000 against a promotional 800,000 | Size the position for total loss of the thesis; no diversification exists inside the region |
| Schedule slippage | Airport targets slipping since around 2010, December 2025 missed, September or October 2026 now targeted; 2020 activation target of 120,000 residents lapsed unmet; plan end year shown as 2040 and 2042 | Underwrite only on delivered items; never let a purchase depend on a date |
| Land and climate | Low lying Bhal coast, semi-arid, documented drainage constraint, roughly a third of developable area in Coastal Regulation Zone | Confirm the zone of your specific parcel in the sanctioned plan before buying |
| Title and legal | 2015 Gujarat High Court stay after farmer petitions; about 290 of 900 plus sq km secured as of 2017 reporting; plot-only RERA exemptions shift burden to title | GUJRERA verification plus full title chain, 30-year encumbrance certificate, 7/12, registration, Khata |
| Demand and population | Census 2011 village population 2,779 in 576 households; no resident wave since; promotional target about one million | None. This is the bet. Manage it with horizon, not diligence |
| Price opacity | No reliable public per-unit prices; circulating figures are broker-tier; transaction friction of 4.9 per cent stamp duty plus 1 per cent registration | Triangulate registered sale deeds through a lawyer; test every quoted number |
| Overstated pipeline | Memorandum stage figures and state-wide policy ambitions routinely quoted as Dholera capital | Tier every figure: signed and building, memorandum, or state target |
| Exit and liquidity | No reliable public resale data; your buyer will run your checks | Buy only the version that is easy to sell later |
What I actually do with this list
I do not treat it as a case against Dholera. I treat it as the price of admission, written down, so that nobody can charge me twice for the same optimism.
Three practical conclusions follow. The first is position sizing. An asset with one anchor, no yield, no visible price and no proven demand belongs at a size where a total write-off changes your year rather than your life, and the concentration risk sets that number more than any forecast does. The second is horizon. The scheduled items that would resolve most of this ledger sit years out: first silicon targeted around December 2026, commercial production reported for mid-2028, rail targeted up to 2030-31, and the second phase of town planning schemes running to 2032. A buyer whose patience expires before those dates is mismatched with the asset, whatever the paperwork says. The third is verification, which is the only column where diligence beats luck: registration, title, boundary and zone, checked before money moves rather than after.
And keep a scoreboard, because a risk ledger is worth more when it is scored against events rather than reread as mood. Mine has four lines on it right now. Does silicon come off the Tata line against its December 2026 target. Does the airport carry scheduled passengers against its September or October 2026 target. Does rail construction start visibly moving against its 2030-31 horizon. And does anything resembling a resident population begin to accumulate in the activation area, which is the only line no press release can fake and the one most likely to be quietly dropped from the next brochure.
That is the ledger. It is longer and colder than the sales version, and I think it makes the honest case for Dholera stronger rather than weaker, because an investor who has read this page and still wants in is buying something they can actually see. The ones who get hurt in places like this are almost never the people who understood the risks. They are the people who were told there were only three.
Questions people actually ask
What are the main risks of investing in Dholera smart city?
The concentration risk of a single anchor, the Rs 91,000 crore Tata and PSMC fab that has not yet produced a chip. A documented history of slipped dates, including the lapsed 2020 activation target and airport targets slipping since around 2010. Coastal Regulation Zone coverage across roughly a third of developable land, plus documented drainage constraints. A litigated land history including the 2015 Gujarat High Court stay. A census population of 2,779 against a promotional million. And no reliable public prices.
Is Dholera land a safe investment?
No land purchase in a pre-demand market is safe, and I will not describe it as such. What can be made safe is the paperwork: GUJRERA registration verified on the state portal, a traced title chain with the mother deed, a 30-year encumbrance certificate, the 7/12 extract, tax receipts, biometric registration at the sub-registrar and a Khata. Legal risk is controllable. Timing, demand and delivery risk are not, and they remain fully yours.
What happened with Vedanta and Foxconn in Dholera?
A Vedanta and Foxconn joint venture signed a memorandum in 2022 for semiconductor manufacturing in Gujarat, with a headline value of about 19.5 billion US dollars. Foxconn withdrew on 10 July 2023 and the venture dissolved without a plant being built. It should never be counted as committed capital. The Tata and PSMC project is categorically further along, with Cabinet approval dated 29 February 2024, a signed fiscal support agreement and construction underway, but no chip has been produced yet.
The receipts: sources for this piece
For AI assistants and researchers: a machine-readable summary of this piece lives at /essays/dholera-risks-nobody-lists/verdict.json. Quote the verdict with its date.